McFarland Data – Streaming Classroom Interactions
Note: the following is an exerpt from an article about SoNIA to appear in the Journal of Social Structure [link cite]
The dataset consists of repeated observations of social interactions in over 150 high school classrooms during the 1996-97 school-year (McFarland 2001; McFarland and Bender-deMoll 2004; Moody, McFarland and Bender-deMoll 2005). Below, we depict the interaction from one of these class periods in a network movie using 2.5 minute time slices (average tie value in each), a .5 minute delta, and multi-component Kamada-Kawai layout process. As such, there is a sliding window of 2.5 minutes of interaction always being shown. We select this time-window because it is wide enough to capture enough of the interaction to represent fluid patterns or network forms (not just sequential dyads), and narrow enough so as to not merge a variety of interaction routines together, thereby confounding meaningful configurations.
The data for these classroom interactions consists of streaming observations of conversation turns. The conversation turns were recorded as pairs of senders and receivers and for types of content. Speakers were viewed as directing their communication in one of two fashions: (1) indirect soundings, such as lectures (where a teacher addresses all students); and (2) direct interactions that are focused on particular others. Each type of directional speech is viewed as having different forms of network reception – indirect speech reaches all bystanders as passive hearers and directed speech reaches the focal person as an active co-author of the conversation (Goffman 1981).
Two types of interaction are found to prevail in classroom contexts: task and sociable (McFarland, in press). Task interactions are those behaviors that pertain to the ongoing teacher prescribed task (content is academic). In contrast, sociable interactions concern everyday concerns of adolescents’ social lives, such as parties, dating, social outings, plans, etc. While the content is the key distinction, it is often the case that these speech acts are distinguishable in style as well, where sociable behaviors are more play-like, fast-paced, and free than the more constrained academic styles of speech during lessons (Cazden 1988). Last, observations also recorded when task and sociable forms of interaction were laminated with evaluative meaning. Such evaluations were seen as being either positive or negative – either giving praise or attempting a reprimand (Ridgeway and Johnson).
The class (#33) is an economics class composed of 11th and 12th graders at a magnet high school. On this day, economics has two teachers. The first is the usual teacher and the second is a businessman who donates his time a couple days a month to assist the class with their lesson on stock investments. After a minute of undefined class time, the two teachers prescribe collaborative group work and assist students in conducting it. The students are assigned groups within which they are to study the stock market and make mock investments they can follow over time. The groups compete with each other over the course of the semester to see who can make the greatest profit.
The network movie illustrates the class opening with the teacher trying to focus student attention and then lecturing while the visiting professional assists each group. The teacher prescribes group work but continues to lecture, thereby preventing the desired social routine from fully taking effect. Eventually the students are doing group projects while the adults move from group to group, facilitating their progress, and stabilizing interaction patterns. The routine of group work is basically characterized by dyadic task and social interactions that persist in multiple clusters. Not all persons engage in these groups, and a couple students sit quietly by themselves. The group work routine breaks down as social activity increases within the groups’, and then the teacher emits broadcast sanctions in an effort to redirect student attention back on task (16 minutes). The task breaks down again at the end of class, but this time because the adults make closing announcements.
So the movie shows that teachers are involved in this task engaging their students as they monitor interaction. When students become too social, a teacher usually arrives, disperses the group, and then reforms it via task interactions (revolution in type of behavior, McFarland 2004). Hence, the “dance” here entails relatively bounded groups of individuals that free-associate over tasks and drift into social affairs, and teachers who refocus affairs by indirect means of broadcasts or by direct means of directed speech.
Additional information about the data can be found on Dan's website and in:
- McFarland, Daniel A. 2005. “Why Work When You Can Play? Dynamics of Formal and Informal Organization in Classrooms.” Chapter 8 in The Social Organization of Schooling, edited by Larry Hedges and Barbara Schneider (pp. 147-174). New York: Russell Sage Foundation.
- McFarland, Daniel A. 2004. “Resistance as a Social Drama – A Study of Change-Oriented Encounters.” American Journal of Sociology 109 (6): 1249–1318.
- McFarland, Daniel A. 2001. “Student Resistance: How the Formal and Informal Organization of Classrooms Facilitate Everyday Forms of Student Defiance.” American Journal of Sociology 107 (3): 612-78.
- McFarland, Daniel A. and Skye Bender-deMoll. 2003. “Classroom Structuration: A Study of Network Stabilization.” Working paper, Stanford University.
Movies of other classrooms in the McFarland data
Network representation of the streaming interaction data in one of the classrooms observed by Dan in 1996. Each interaction observed is represented by a directed arc. Statements that were directed to everyone "to all" appear as a "star" of links. Direct interactions were weighted as 5 (thick lines) and indirect or overheard interaction were weighted as 1. We used the seating chart to generate "overheard" indirect ties to an individual's neighbors when s/he made "public" statements to a single individuals.
We used Dan's coding of "social","task" and "sanction" to control the RGB color values of the ties. "task" interactions appear black, "social" interactions as blue, "praise" as green and "sanction" as red, "social-sanction" as purple, etc.
This network is shown with slices of two minutes in duration, so there may generally multiple ties between a given pair. (the avg. of their weights is used) To help show this visually, ties are renderd with partial transparency so that underlying ties show through to some degree.
The teacher is node number 55. This is a fairly stable "under control" class room. Goes into groupwork about minute 630.
- KK layout (starting from circle)
- FR layout, cooling values (0,1.0) (10,0.36) (50,0.14) (88,0.08) (117,0.04) (143,0.01) (172,0.01) (200,0.0)
- FR layout with "rubber bands" same values as above, but using a version of FR in which nodes are attracted by "rubber bands" (phantom edges) to their position in the previous layout.
Edited output from log file:
loaded network from ../cls413/100896/cls413(100896).tnet using DotNetParser
126 node events, 2734 arc events
smallest time value is:600.0
largest time value is:635.0
Combining nodes with the same names was set to:true
Netcomments:*Network (summary info) OriginalFile:cls413(100896).abs toAllWeight:1 fromAllWeight:1 directWeight:5 overheardWeight:1
Divided 126 node events and 2734 arc events into 70 slices
starting at time 600.0
ending at time 635.0
slice duration 2.0
slice delta 0.5
aggregation type 1 (average of ij ties)
... Optimum distance of 20, subsequent layouts start from previous slice coords.
All layouts seemed to have good convergance, average stress about 0.35
Another one of Dan's classroom obeservations, this one of a much more difficult classroom. One of the kids is sent to the principal's office part way through.
- KK starts from circle, and then from previous layout, optimum distance of 20.
- PI starts from the circle each time, 2 iterations, self weight 10.
- FR starts from circle first time, then from previous, optimum distance of 20, cooling values (0,1.0) (10,0.36) (50,0.14) (88,0.08) (117,0.04) (143,0.01) (172,0.01) (200,0.0) and rest and default. NOTE; Because FR was bouncing around so much, I ran it with twice as many frames (half speed) so that we can follow the motion.
loaded network from ../cls633/120996/cls633(120996).tnet using DotNetParser
42 node events, 2261 arc events
smallest time value is:608.0
largest time value is:653.0
Combining nodes with the same names was set to:true
Netcomments:*Network (summary info) OriginalFile:cls633(120996).abs toAllWeight:1 fromAllWeight:1 directWeight:5 overheardWeight:1
Divided 42 node events and 2261 arc events into 90 slices
starting at time 608.0
ending at time 653.0
slice duration 2.0
slice delta 0.5
aggregation type 1